LINpro &

LINsurround
User Manual

MAAT Inc.
 MAAT Incorporated
101 Cooper St
Santa Cruz CA 95060 USA

More unique and essential tools and tips at:

www.maat.digital
 Table of Contents
Installation & Setup.................................................................................. 5
Licensing.. ................................................................................................................................... 5

Online Activation..........................................................................................................................................................................6

Offline Activation.........................................................................................................................................................................6
Introduction. . ........................................................................................... 8
Quick Start............................................................................................... 8
Dithering Basics. . ...................................................................................... 8
Why Shape?................................................................................................................................ 9

Shaping The Noise................................................................................................................... 10

The Interface. . ........................................................................................ 10

Dither Details......................................................................................... 11
Shaped Dither...........................................................................................................................12

Limitations............................................................................................ 13
In Use.................................................................................................... 14
Optimized Choices....................................................................................................................14

New To Dither?......................................................................................................................... 15

Choice Is Good.. ........................................................................................................................ 15

Output Bit Depth........................................................................................................................................................................16

Bit Depth Reduction Mode.......................................................................................................................................................16

Dither Type...................................................................................................................................................................................17

Noise Shaping Type....................................................................................................................................................................17

Noise Shaping Enable/Disable.................................................................................................................................................18

Auto–Blanking.............................................................................................................................................................................18
Delivering From Floats To 24 Bit Fixed.. .................................................................................. 19

Preferences. . ............................................................................................................................. 19

Presets.................................................................................................. 19
Specifications. . ....................................................................................... 20
System Requirements.............................................................................................................. 20

Supported formats. . ................................................................................................................. 20

Updates................................................................................................. 20
Support................................................................................................. 20
License Central......................................................................................................................... 21

In Use.............................................................................................................................................................................................21

Controls.........................................................................................................................................................................................21

Preferences.................................................................................................................................................................................. 22
Share The Love.. ...................................................................................... 23
Support................................................................................................. 23
Credits . . ................................................................................................. 23
Appendix One — The Math.. ..................................................................... 24
Appendix Two — The Curves.................................................................... 24
Appendix Three — The Curves.................................................................. 25
About This Manual.................................................................................. 37
Installation & Setup
MAAT LINpro and LINsurround are based on the same plug–in. A
mono/stereo license enables LINpro, while a multichannel license
enables LINsurround. When preparing for installation of your MAAT
LINpro and LINsurround, we recommend that you quit all applications
prior to proceeding. Also, after downloading the installer and prior to
installation, let any anti–malware measures you may have running scan
the installer. Then, temporarily disable all anti–malware measures once
scanning is complete.
To install, simply double click on the downloaded Installer for your
particular operating system. The installation process will guide you
through the install procedure. If you have an older OS version, restart-
ing your host may be required. Don’t forget to reënable anti–malware
measures or simply restart your host.

Licensing
MAAT software is licensed through our dongle–free, cloud–based
license control framework. If your hosts are net–connected, our licenses
“float,” making it easy to move from one machine to another. All you
have to do is close all MAAT stuff on one machine before you open
anything on another. If you’re going to use the laptop in an environment
where Internet is spotty or non–existent, you can also go offline tempo-
rarily for up to 30 days.
To license your MAAT purchase, open yours or instantiate an example
in your preferred plug–in host. When licensing, we recommend you
open a new session in your DAW just for this purpose.

Figure 1: The initial Activation dialog

Page 5 of 37
Online Activation
Once the Activation dialog appears, paste or type in your Product Key
supplied at the time of purchase. Then click on Activate/Deactivate On-
line and follow the prompts.

Z NOTE: The Online process requires a live internet connection. You must
have an active internet connection for activation.

A feature of our cloud–connected system is that, if you close a DAW

session that contains MAAT plug instances, then open that session on
another host, the license will “follow” you onto the new host as long as
you have live Internet.
You can also return or park your license on our licensing server, and
pick it up later on another host. Once your product is licensed, click
the Activate/Deactivate Online button to deactivate your license, then
repeat to reactivate.

Offline Activation
If, for security reasons you have an offline host computer or you plan on
working without a reliable Internet connection, it is best to go with an
offline license. There are two offline options, temporary offline and full
offline.

Temporary Offline
If you plan on working without a reliable Internet connection, we
designed the temporary offline option just for you. The temporary
offline process also requires a live Internet connection, but only
during activation.
A temp offline license has two unique features:
• It can remain offline, without an Internet connection, for up to
30 days
• It auto–renews its lease whenever it does make a server connec-
tion
So, if your Internet is spotty or unreliable, or you’re heading tem-
porarily to a location where an Internet connection isn’t available,
the temporary offline option let’s you work untethered to the ’net
for up to a month.
By clicking the “30 Days Offline” button, your online license will
become temporarily offline for a maximum of 30 days. If, however,
you use the product while connected to the Internet at all during
that time, the “lease” duration is automatically reset, extending an-
other 30 days.

Page 6 of 37
 Figure 2: The 30 Days Offline activation option

Full Offline
This will license your MAAT product without the need to periodi-
cally contact our licensing server. If you anticipate working without
the Internet for more than a month at a time or, for security purposes,
your studio machine is “air gapped” and has no Internet, take a
moment to surrender your online license, and replace it with an
offline version.

Z NOTE: a full OffliNE licENsE dOEs NEEd TO bE EXpliciTlY rENEWEd ONcE

a YEar.

To prepare for offline activation, you will need:

1. Your Product Key!

2. Some method to move your License Confirmation

file to the offline host machine.
That latter requirement can be accomplished in any number of
ways, but a USB flash drive is often readily available and most con-
venient. You will also need your Product Key that was supplied to
you via e–mail when you purchased your MAAT product.
The offline activation process starts with generating a License Re-
quest, with a “maatr” file extension, which is submitted on–line

Page 7 of 37
 to our server or via e–mail to MAAT Support <support@maatinc.
zohodesk.com>. A License Confirmation file, with a “maatc” exten-
sion, will be returned to you, which will activate your product.
To proceed, click on the Activate/Deactivate Offline button, and
follow the prompts.
In a year’s time, 365 days from the date the maatr file is created, a full
offline license must be renewed using the same maatr/maatc method.

Introduction
Thanks for purchasing MAAT’s LINpro re-dithering plug–in, and
congratulations on committing yourself to maintaining and preserving
linearity in music!
We built LINpro and LINsurround to be used during post–production,
when redithering is most needed. This is especially true during the mas-
tering stage, where it is quite common to play back a digital stream and
send that to one or more analog devices. Once the signal passes through
the analog equipment, it is re–digitized and further processed in the
digital domain. These steps require re-dithering and we provide quite a
bit of control over that process so the engineer can either use the opti-
mal dither for your release format, or a custom setting for a subjective
“sound.”
Today, all audio workstations allow high quality production processes:
recording, mixing and mastering, normally with at least 24–bit final
output quality. However, with the exception of iTunes and a few other
services, lossless distribution and storage format is often the 16–bit lin-
ear PCM format widely used for CD production and lower quality loss-
less streaming. If optimally achieved, 16 bits are sufficient to deliver an
excellent quality end product, far better than any lossy coding format.

Quick Start
We know, reading is old school and such an easy tool to use as LINpro
would seem to be self evident, wouldn’t you think? However, before you
begin your explorations with this tool, we want to make a few things
clear so you will get the best use out of your “LIN” purchase. Thanks for
struggling through this part!

Dithering Basics
Before we dive into details, here are a few very important concepts to
remember:
• Dither prepares the finished audio for down–stream processing
by the consumer
• When in doubt, TPDF is the universal default choice
• DACs usually do not dither
We will elaborate more shortly but, when in doubt, use High Passed Tri-
angular Probability Density Function or HP–TPDF dither. It will func-
Page 8 of 37
 tion properly in all delivery and playback scenarios. Note that DACs
or Digital–to–Analog Converters usually do not redither. So, for DACs
with digital volume controls, the audio quality will not be optimal when
the volume control is used.
To produce 16–bit formatted files, high resolution source audio files,
minimally 32–bit floating point or 24–bit fixed, have to be reduced to
a 16–bit word length. After this operation however, referred to as “re–
quantization,” a whole range of problems arise:
- the information beyond the 16th bit is lost — fine, low-ampli-
tude audio detail beyond a 16–bit quantizing level disappears
- the noise floor increases to a 16–bit quantizing level
- for large amplitude signals, harmonic and intermodulation dis-
tortions appears
- for small signals, nonlinear distortions rise dramatically — small
signals are precisely where modern analog devices are generally
linear
- for small signals, stereo image instability appears
- at normal levels, unnatural falling off of reverberation occurs —
smoothly fading into the noise floor becomes impossible
Especially for low level signals, the quantization errors correlated with
and modulated by the original signal can be quite noticeable. Re–
dithering technology helps to decorrelate the quantization error and
consequently convert these discrete, unpleasant distortions into white
spectrum noise. Proper re–dithering reduces or eliminates many of the
problems listed above.
The decorrelation process is performed by adding a random signal to
the high resolution audio signal prior to the re–quantization process.
This random signal is called dither. Dither performs linearization of the
nonlinear characteristic of the quantizer.
Re-quantized and dithered content sounds more pleasant, one can say
“more analog.” The price of this improvement is a noticeably higher
background noise level. However, it is white in character and constant
in level. Especially at low signal amplitudes, where the error is correlat-
ed to input signal, quantization noise sounds subjectively much more
unpleasant than the even higher amplitude uncorrelated white noise
produced by proper dithering.
The remarkable effect of a properly implemented dither algorithm is that
a 16–bit signal can clearly reproduce tones at -110dB. After re–quanti-
zation without dither, there is simply no output for input signals below
-96dBFS. All low amplitude detail is lost.

Why Shape?
However, with only dithering, we still have no influence on the level and
audibility of the quantization noise. Therefore, in addition to dithering,
Page 9 of 37
 we use noise shaping algorithms. Noise shaping modifies the whole
distortion spectrum: re–quantization plus dither noise. The noise power
in the system cannot be removed but it can be distributed so the noise
after shaping is significantly less audible. It is even possible to partially
achieve an SNR or signal–to–noise ration below the original quantiza-
tion noise floor. It is important to mention that the input signal is trans-
mitted transparently through properly implemented dithering and noise
shaping processes up to the re–quantization point.
In general, it is possible to use noise shaping without dither but a sig-
nificant correlation between successive rounding error samples occurs;
distortion and signal–dependent noise modulation. Therefore, we rec-
ommend always using noise shapers with proper triangular dither.

Shaping The Noise

Because the ear’s sensitivity to low amplitude broadband noise is not
uniform with frequency, we use noise “weighting” curves. Weighting, as
mentioned above, redistributes the power spectrum of the noise. Dith-
ered re-quantizing noise shapers approximate the inverse low amplitude
noise audibility curve (in the old days, audibility curves were referred to
as “Fletcher–Munson curves”) allowing one to achieve the least audi-
ble noise penalty. Again, this leads to the counterintuitive conclusion
that psychoacoustically optimized noise shapers allow for 19–bit audio
performance from a 16–bit standard media delivery format. Signals with
amplitudes as low as -120dB can be clearly detected.

The Interface

Figure 3: LINpro’s user interface

Page 10 of 37
 The straightforward user interface contains a series of menus and pa-
rameter controls. All of LINpro’s functionality is wrapped in a visually
unobtrusive user interface. The plug–in is also very “light weight,” de-
manding a minimum of CPU resources so it won’t slow down your host.
We’ve worked hard to make this an exceptionally useful and usable tool
and we hope you’ll dig it.

Dither Details
In the digital audio world, dither is noise added to reduce distortion.
What? Adding noise to decrease distortion?! Yup, let’s dig in some
more…
Dither is an interesting and fundamental aspect of digital audio. Al-
though not usually thought of as “DSP,” re-dithering is definitely signal
processing and so should be part of the overarching DSP umbrella. Back
in the Distant Past, re-dithering was an integral part of our original
digital telephone system, and much of the pioneering research on the
subject was performed at Bell Labs.
According to information theory, you should add a pinch of noise to
your digital audio every time you change it. Why is that? Well, take a
look again at figure 3 below; that blocky looking mess that used to be a
smooth and continuous low amplitude sine wave.

Figure 4: A digitized sine wave

Because a digitizer can only choose one discreet sample value during
each sample (time) period, not half of one and half of another, adding
a touch of noise helps to nudge the system to land on the statistically
“right” choice. Injecting a hint of noise “linearizes” a digital conversion,
reducing distortion. A properly designed digital system needs to add
noise, called re-dithering, every time you change the gain (multiplica-
tion), or mix signals together (addition). That includes EQ, effects and
pretty much anything else we typically do to our client’s music!
The ideal amount of dither noise is about half of the step size of the
digitizer, though it differs depending on the type of noise being used. In
the case of good ol’ linear PCM on which most all of our music is based,
is 0.5 times 6dB or 3dB is correct for rectangular dither (RPDF). For
triangular PDF, the ideal is about 4.7dB. Either way, half a bit is a very
small amount of additional noise.

Page 11 of 37
 Without proper re-dithering, the distortion that results from any add or
multiply is “correlated.” This means it bears an audible relationship or
subjective connection to the signal, the music. Trouble is, it’s distortion,
so it’s a nasty sounding parody of the music and is usually considered to
be quite unpleasant to listen to. When properly re-dithered, the signal is
linearized, resulting in the ability to hear clean music within the noise
floor at less than one bit of amplitude. In other words, that LSB, or Least
Significant Bit, now can contain meaningful music “under” the noise.
There are several common types of dither, with the simplest being
broadband noise, known as rectangular probability density function
(RPDF) dither. Mathematically, this RPDF dither has uniform or equal
probability of including any frequency within the audio band from 20 to
20 kHz. So, it has equal power at all frequencies and sounds “white.” Be-
cause of its power distribution, it is quite audible, even at low amplitude
and so should be used with caution. More common is triangular PDF or
TPDF, which has less low frequencies and more high frequency energy.
Because our hearing is less acute at higher frequencies, TPDF dither is
generally less audible than RPDF dither.
As mentioned earlier, dither can linearize a digital audio system, sig-
nificantly reducing distortion for low amplitude signals, and increasing
the signal resolution well below the quantizer step size. Dither does add
some additional noise. However, even if the wideband noise has about
the same total power as the signal, the high Q filtering action of the
basilar membrane allows our ear to clearly resolve sine waves well below,
“buried” in the noise.

Shaped Dither
Back to shaped dither…A fairly modern invention is “shaped” dith-
er. Shaping refers to the power spectrum of the dither signal, which is
tailored or shaped for even less audibility. Noise shaping is a technolo-
gy which reduces the audibility of noise resulting from dithering and
re-quantizing. Noise shaping is accomplished not with equalization, but
by processing the signal created by comparing the high resolution input
and re–quantized output.
Our hearing mechanism is optimized for the frequencies around which
the human voice is centered so, higher frequency sounds are less audible
than lower frequencies. The more you shape the dither noise, shoving
the dither energy toward higher and higher frequencies, the less audible
it becomes…sort of. When used in larger than normal quantities, as
with several generations of processing, or when set to extreme shaping,
the dither signal itself becomes obnoxious, and is a bit like the sound of
nails on a blackboard.
Noise shapers can modify the distortion spectrum after re–quantization
in a way such that the signal–to–noise ratio can be improved over the
theoretical quantization noise floor. The noise floor can be optimized to
allow for the least possible audible noise penalty in a system. To quote
a 1992 AES paper by Robert A. Wannamaker, the “…object (of shaped

Page 12 of 37
 dither) is to minimize the total perceived output noise power. For this
purpose, a new perceptual frequency-weighting function is introduced
which provides closer matching to the ear’s measured sensitivity at high
frequencies…” Noise shaping shifts the noise energy from high audibil-
ity regions where the ear sensitivity is highest to areas with lower sensi-
tivity.
There are a variety of different shapes, from simple to very sophisticated,
all based on psychoacoustics. An advanced, ninth order noise shaper
can reduce the perceived output noise power by about 17 decibels. With
some noise shaping schemes, performance equivalent to 19 bits of preci-
sion are possible within a 16–bit delivery format.

Z NOTE: Even if you decide to use un–shaped dither for your final deliver-
able, shaped dither is useful for monitoring prior to delivery.

Since shaped dither moves the subjective dither noise away from our
most sensitive hearing region, you can used shaped dither during post–
production to best perceive the subtleties of the music. Once all post and
mastering is finalized and the content is ready for delivery, then you can
choose the best dither method for the distribution formats and scenari-
os.
Some examples of shaped dither brands are Apogee’s UV22HR, iZo-
tope’s MBIT+, the Power Consortium’s POW-r, and Sony’s SBM II,
(Super Bit Mapping Two). MBIT+ and POW-r are adjustable, which is
useful only when applying dither as an effect. The above mentioned
branded redithering schemes also require a license, increasing cost. We
have evaluated and improved on those legacy methods for LINpro.

Limitations
When processing is performed in a DAW, dithering should be applied at
the very end of the deliver chain. In actuality, this is re–dithering since
the ADC or analog–to–digital converter self–dithers the signal during
initial digitization.
Notice I said “at the end…” This is important in that dither should only
be re–applied when the content is moved out of your DAW, either by an
AES/EBU or other streaming spigot, or as a file or set of stems.
In the 1990s, manufacturers began developing re-dithering schemes
that provided the linearization without the audible addition of noise.
The first of the resulting products was Apogee’s original UV-22, men-
tioned above, an effective if brute force approach. A handful of better
performing and better sounding spectrally shaped re-dithering products
followed from several manufacturers.
Unlike the early days of digital audio, many, if not most consumers now
listen to music through a largely digital signal chain. The DAC may be
in their phone, embedded in their headphone cable or inside the “smart
speaker” on the kitchen table. As legendary mastering engineer Bob

Page 13 of 37
 Olhsson mentioned when asked about his re-dithering choices, “In
most cases, people are never going to hear (my product) without some
kind of downstream digital processing.” Theses days, only audiophiles
have integrated amps or stand–alone preamps with analog gain staging.
Re-dithering with anything other than TPDF does not “translate” well
in our modern, all–digital world, as shaped dither increases distortion
rather than reducing it.
With that fact in mind, one can also think of dither as just another
effect, a “color” paintbrush in the DSP toolbox. No re-dithering at all
produces a crunchy aspect that one well known mastering engineer
employs as part of his signature sound, while more that a half step adds
a veil of extra noise that may be perfect for your particular client.
Our point is: rely more on your ears than your eyes. Take some time to
listen to the different dither choices available in LIN, and learn what
they do to a piece of reference music played through reference consumer
systems. It’s easy to do…grab some audio with which you’re familiar,
and gain adjust it down by 8 bits or more, that’s at least 48 decibels.
Then, instantiate LINpro and turn your monitor gain up enough to hear
what the different settings are doing.
To make your dither choices very audible, increase the amplitude. This
is easily done by reducing the Output Bit Depth setting. Since 8–bit
digital audio holds only 48dB of dynamic range, LINpro’s output at that
setting is around 48dB below full scale or 0dBFS. We also give you an
8–bit output word length setting to help you differentiate between the
effect each setting has on your material.
Once you’re familiar with the setup and the character of the dither gen-
erator, try a much larger gain reduction so you are “down in the mud,”
where dither really counts. Be careful as this requires a huge amount
of monitor gain or, use LINpro’s 8–bit mode. Then, compare TPDF to
rectangular and noise–shaped in an all-digital playback scenario com-
pared to that much rarer all–analog playback situation. Remember to
gain-reduce the digital playback chain so the playback engine processes
the signal one more time. Then, level match the analog chain and start
comparing.

In Use
Optimized Choices
The collection of controls include output bit depth, bit depth reduction
method, dither type and noise shaping type. Noise shaping bypass and
auto–blanking round out the settings choices.
In general, dither and noise shaping types can be chosen in any combi-
nation, but not all combinations are recommended. To avoid sub–opti-
mal results, we only allow triangular and triangular high-pass dither in
combination with any noise shaping type. For comparison purposes, a
no–dither choice is available. In this case, rounding only at the request-

Page 14 of 37
 ed bit width is applied. Truncation, which causes significant re–quanti-
zation distortion and DC offset, is included for educational purposes.

New To Dither?
8-bit mode is also implemented for training purposes, to provide easily
perceived experimentation with dither types and noise shaping choices.
Dithering to 24 bits is useful when moving from 64–bit floating point
to fixed point. It also helps with 32–bit floats, but only just. The 32–bit
floating point format has only a 24–bit mantissa, and therefore there’s
not that much information to preserve.

Z NOTE: When not using noise shaping, the use of triangular high-pass
dither is a recommended practice.

If all this dither madness is new to you, we suggest you start out in 8 bit
mode, with dither and noise shaping disabled. Set up a project just for
training your ears, and add several tracks of different genres. In your
DAW’s Master section, reduce the gain of your clips by 48dB. Then, in-
sert LINpro and, downstream from that, insert/dial up gain makeup of
+48dB. All that will allow you to better hear the result of your LINpro
manipulations…
Again, start out in 8 bit mode, with dither and noise shaping disabled.
Listen to how the program material “sounds.” Get used to that, then
enable truncation and again listen to how that sounds. Now, switch
from truncation to rounding. Note that dither is still disabled…Once
you have a solid sense of how those different settings affect sound qual-
ity, enable the simplest possible dither setting; TPDF with no shaping.
Compare that effect to high passed TPDF. Next, compare the various
shaped noise options…With all that aural experience, you are now ready
to “dive into the deep end”!

Choice Is Good
All of LINpro’s psychoacoustic noise shapers have been precisely opti-
mized for the 16 bit/44.1 kHz PCM format, also known as “Red Book”.
They profit from inherently lower quantization noise in the audible fre-
quency range, and are also applicable for 48 kHz deliverables. Of course,
dithering and 1st order noise shaping work perfectly with any sample
rate and any playback chain, analog or digital. For deliverables at 88.2
kHz and higher sample rates, use the default TPDF dither with 1st order
noise shaping.

Output Bit Depth

Dither Type Shaping Comments
Format Reduction

8 bit No Change None None

16 bit Round TPDF NS_1 1st Order Noise Shaping

24 bit Truncate HP–TPDF NS_2 2nd Order Noise Shaping

RPDF ‡ NS_3 3rd Order Noise Shaping

Page 15 of 37
 PS_1 Psychoacoustic 1

PS_2 Psychoacoustic 2

PS_3 Psychoacoustic 3

‡ – Psycho- PS_4 Psychoacoustic 4

acoustic noise
shaping is not
available

Table 1 : Setting choices

Output Bit Depth

For your deliverable word length, 8, 16 or 24 bits are available. An 8 bit
setting provides a very obvious audible result, making it easy to arrive at
an optimal setting for your content.

Bit Depth Reduction Mode

In the words of our Head of Development, this setting, “reaches into the
bits of every float and deciding what to do with them.” Effective preci-
sion determines what the really least significant bits contain. If the Out-
put Bit Depth is set to any value other than 24, then this menu decides
the contents of the defined LSB and all bit cells past that.
The choices are Round or Truncate. Of those settings, only Round(ing)
will produce optimal results. Truncate will result in loss of all informa-
tion in the lower order bits. Rounding without dither is not recommend-
ed because the lowest amplitude information is lost.
The Round setting rounds to the desired word length. At this setting,
the specified LSB represents a single bit rounding of all the lesser bits
past it, those carrying information about the signal that are lower in
amplitude. If Output Bit Depth is set to 16, then the 16th bit will contain
a value that results from rounding the 17th through 24th bit. The 17th
through 24th bit will then be “padded” with zeros or “null data.” Round
is the default as rounding the excess bits produces the most linear result.
The Truncate setting simply “chops off” or truncates all bits past the
LSB for the specified word length. If Output Bit Depth is set to 16, then
the data contained in the 17th through 24th bit is discarded, and those
bits are then filled with null data (zeros).
For a slightly “crunchy” sound, use the Truncate setting in conjunc-
tion with Output Bit Depth set to 16, plus TPDF type dither and zero
dither gain. The zero gain setting means no dither noise will be ap-
plied and, because of truncation, LINpro will simply toss out the 17th
through 24th bit. This means the 16th bit will not contain any low
amplitude information, as it would if dither had been injected. With
dither, information in the 17th through 24th bit would have pre-
served. The crunchiness is caused by the low amplitude loss of detail
generated when truncation is employed.

Page 16 of 37
Dither Type
Four choices: that include Rectangular or Triangular Probability Densi-
ty (RPDF/TPDF), plus none or High Passed TPDF.
NONE simple truncation, no dither; not recommended because
of asymmetrical quantization error
TPDF dither with triangular probability; spectrally flat, addi-
tional 1.8dB noise floor, but no noise modulation, highly
recommended especially with noise shaping
HP–TPDF dither with triangular probability and noise shifted to-
wards high frequencies; results in reduction of audibility,
universally applicable, highly recommended especially
when noise shaping is not used
RPDF dither with rectangular probability; historical, not recom-
mended because the 3dB higher noise floor is modulated
by the input signal

Z NOTE: When dither is disabled or set to NONE, then noise shaping (see
below) is also disabled.

Noise Shaping Type

Shaped noise is not broadband noise that has been EQ’d. Shaped noise is
derived from the method used to generate the noise, not how the noise is
post–processed. Unlike plain TPD noise, which affords you an extra bit’s
worth of signal underneath the noise floor, shaped noise can add two
bits or more of perceptually clean signal embedded in the noise floor.
NONE No noise shaping; resulting noise spectrum is more audi-
ble than a shaped dither.
NS1 First order, non–modified error feedback; results in a con-
tinuous shift of noise to higher frequencies, allows percep-
tual preservation below the theoretical quantization noise
floor for low frequency range (up to ≈3 kHz) with a slight
increase in noise level at higher frequencies where audibil-
ity is much lower
Starting with 2nd order error filtering, we enter the range of weight-
ed noise shapers. They allow adjusting noise according to modern
psychoacoustic research. Audible noise can be reduced extensively by
about 17dB, almost 3 more bits. After re–quantization of a high resolu-
tion signal, perceptual audio quality equivalent to 19 bits can be re-
tained within only 16 bits of word length.
The most common curve in psychoacoustics is A–weighting. How-
ever, this curve is not satisfactory to approximate the low amplitude
sensitivity of human hearing. Since dither is applied to optimize the
LSB or Least Significant Bit, a better choice is to employ E–weighting,

Page 17 of 37
which is based on the inverse of the ISO 226 15 phon equal loudness
curve modified for diffuse field.
NS2 2nd order error noise shaping; approximates the modified
E–weighting curve; offers almost two bits of improvement
in apparent SNR, in the range of approximately 3-4 kHz,
the highest loudness sensitivity
NS3 3rd order error noise shaping; a closer approximation to
the E–weighting curve then 2nd order
Psychoacoustically optimized noise shapers have been designed to
modify the noise spectrum to be least audible across the entire audio
spectrum.
PS1 Psychoacoustically optimized 5th order error noise shap-
ing; follows an improved E–weighting curve based on a 15
phon equal loudness curve but with even better approx-
imation of the low and high frequencies then is possible
with the modified E–weighting curve
PS2 Psychoacoustically optimized 9th order error noise
shaping; based on the F-weighting curve, PS2 is a further
refinement of the modified E-weighting curve, perfectly
fitting splined ISO data in the range of 40 Hz to 16 kHz.
This curve can provide up to 17dB of perceived improve-
ment relative to an unshaped spectrum, almost 3 more
bits in the SNR
PS3 Psychoacoustically optimized 8th order error shaping;
newer generation based on the ITU 486 weighting curve
with modified response above 12 kHz according to recent
observation on the ear’s sensitivity to high frequencies
PS4 Psychoacoustically optimized 8th order error shaping;
newest generation based on sensitivity to noise than to
pure continuous tones, this curve is referred to as uni-
formly-exciting noise (UEN) at threshold which describes
the minimum audible SPL of critical–band–wide noise on
frequency axis. The curve was further modified to rep-
resent the power spectrum of UEN at threshold in a free
field

Noise Shaping Enable/Disable

This button disables noise shaping for comparison purposes. Instead of
using the Noise Shaping Type menu, you can turn noise shaping on or
off to instantly hear the effect.

Auto–Blanking
The blanking option, enabled by default, automatically mutes the dither
generator after a preset 11 milliseconds of silence. This prevents you from
injecting noise into what would otherwise be “digital black” or null data.

Page 18 of 37
Delivering From Floats To 24 Bit Fixed
What settings are best when delivering from single or double precision
floats to 24 bit fixed? Noise shaping typically is used to perceptually
“push” the noise floor below what/where it would be with dither alone.
As mentioned previously, noise shaping with dither at the 16th bit can
move the effective floor lower, over the entire audio passband, by up to
~3 bits. So what does that do for you if you’re dithering at the 24th bit?
24 bits encodes 144 dB of dynamic range, which is well beyond either
audibility or reproducibility on normal playback gear at normal play-
back volumes.
With HRA releases, if the data is just truncated at 24 bits rather than
properly dithered to 24 bits, correlated noise will result. Vicki R.
Melchior, current chair of the AES High Resolution Audio Technical
Committee, states that there is “…studio evidence that the truncation
noise can become audible because it will be correlated to the signal. It’s
very hard to show with listening tests, but multiple mastering engineers
have heard it.” Melchior recommends rounding to 24 bits, adding HP-
TPD dither at the 24 bit but not bothering with noise shaping.

Preferences
The gear icon in the upper right corner opens the General pane and
Info pane. The General pane contains one button that, when clicked,
saves the current plug–in state as the default state going forward. The
Info pane contains credits, an Activate/Deactivate License button, and a
button to open this user manual.

Presets
To the left of the Preferences gear icon is a User Preset menu and a right
flanking Save button. The Save button allows you to store a state, while
the Preset menu allows you to recall any factory or user–saved settings.
If an existing preset is modified, an asterisk appears after its name.
Holding down the option or alt key switches the Save button to a Delete
or remove (preset) button. Simply select a preset from the menu. Then,
holding down the modifier key, click the button.
Preset A: Recommended for audio material of average dynamic
range; converts re-quantization noise to subjectively more
pleasant “analog” noise
Preset B: Enhanced by smooth wideband noise shaping with
greatest effect at lower frequencies; works properly for any
audio formats
Preset C: Enhanced by a complex psychoacoustically optimized
curve (tones hearing threshold); recommended for
re-quantization of complex program to the 16-bit Red
Book Compact Disc format; can subjectively retain up to
three additional bits of original within a 16-bit deliverable
Preset D: Enhanced by a new generation, complex psychoacous-
Page 19 of 37
 tically optimized curve (noise hearing threshold); rec-
ommended for re-quantization of high resolution 24-bit
material to the 16-bit Red Book format; can subjectively
retain up to three additional bits of original within a 16-
bit deliverable, greatly increasing perception of very low
amplitude detail, reverberation and spaciousness
LINpro presets are stored as cross–platform XML files. For migrating
presets from one host to another, you’ll find them at…
macOS /Applications/MAAT/LINpro/Presets
Windows C:\Program Files\MAAT\LINpro\Presets

Specifications
System Requirements
• macOS: 10.9 and newer, 64 bit only
• Windows: 7 and newer, 32 & 64 bit
• 4GB RAM minimum

Supported formats
• 44.1 to 384 kHz sample rate
• AAX, AU, VST 2, VST 3

Updates
Please always use the latest version of the software! You can find your
current version on the Info Tab of the back panel. You can download the
latest version simply by visiting:
http://maat.digital/support/#installers
For optimal security and stability, you should always stay up to date
with Operating System revisions, and we keep up with compatibility
changes to our products. We also continue to optimize for reduced CPU
load, and this very user manual gets its own improvements.
To stay up to date with the latest version and product releases, please
subscribe to our occasional newsletter. You’ll find a opt–in form on our
Contact page, or sign up for a copy of our free and very handy 2BusCon-
trol plug–in which will also subscribe you. Don’t worry, we know you
are busy so we only send out an average of 10 or 12 mailings a year.

Support
For product support, please visit:
https://www.maat.digital/support

Page 20 of 37
License Central
License Central, located in MAAT’s Shared directory, is a free utility
that validates, repairs and logs your MAAT licenses. It also displays
what MAAT products you have installed and assists in downloading
both updates and demos.
License Central lists all of our products, and shows you:
• If a product is installed and what licenses you have [CLOUD,
SUBSCRIPTION, TRIAL, OFFLINE and TEMP OFFLINE]
• What versions are installed, and version installers are available
for download
It also provides:
• A one–click download of an update or a demo
• A button to activate or deactivate any of your licenses
• A copyable list of all your Product Keys
• One–click generation of the MAAT diagnostic report
It also fixes license issues automagically.

In Use
When you launch License Central, it scans your host computer and then
attempts to match up each product found with a Product Key on our
license server. That scanning happens in real time, and the process is
shown as a progress bar. When quitting, this process happens in reverse.
Once all the licenses are validated, a list on the left displays all MAAT
products, including License Central. Products that are not installed are
grayed out, and installed products are displayed in high contrast. To the
right of each product entry is a check mark, which indicated that the
product is licensed.
Selecting a product entry displays information about the installed ver-
sion and if an update is available for download.

Controls
There are two persistent buttons along the top. At upper left is Check
for Updates, which refreshes the list of installed products, and pings our
server again for currently available versions to download. It also recre-
ates your Product Key list.
At upper right, the Activate/Deactivate button brings up the familiar
blue and orange MAAT license window, allowing you to deactivate or
return your license to the Cloud, and to switch to a 30 day temporary
offline license. If you have requested a 365 day full offline license, use
the Create License Request function in the MAAT license window, and
have received your “.maatc” confirmation file, you can also take your

Page 21 of 37
license offline. Finally, you can also paste in a Product Key to activate a
new license.

Z NOTE: Switching to 30 day temp offline requires that your license be already
activated. 365 day full offline licenses require a license confirmation file as
supplied by MAAT’s support department.

If any product is out of date, a third Download Update button will ap-
pear when that product is selected. This button is an express method for
updating you MAAT products.
At bottom left of the License Central window is an Auto–Refresh check
box, which is selected by default. Deselecting that will prevent License
Central from frequently scanning your host for changes including new
installs. It starts a scan every 3 seconds, waiting for completion of that
scan, then begins a new 3 second countdown before starting the next
scan. If you find that License Central is interrupting you workflow while
scanning, then disable Auto–Refresh.

Preferences
In License Central, the Preferences’ gear icon brings up the Preferences
window with four tabs.

Paths
For those who set up alternate directories for the management
of their plug–ins in a DAW, two alternate plug–in paths can be de-
fined. These directories will be also included when License Central
searches.

Keys
This tab lists all of your Product Keys that are known to our license
server. Please take a moment to copy all of your Product Keys, paste
them into a text or word processor file, and print out a hard copy.
Your Product Keys are your proof of purchase and, without them,
your purchase will not work and we cannot provide product sup-
port.

Troubleshooting
The Troubleshooting tab has one button, which generates a Diagnostics
Report. This report, automatically written to the Desktop, gath-
ers useful, non–personal information about your machine’s state
that our support team can use to diagnose any problems.

Info
As with all our products, the Info tab lists credits, copyright infor-
mation, and the version number you are running.

Page 22 of 37
Share The Love
Would you like to help us in our quest for better sound quality? Help
support MAAT by sharing the love…Like us on Facebook!
https://www.facebook.com/maatdigital/
and Twitter too:
https://twitter.com/maat_digital
Please tell your friends and colleagues about us. We really appreciate it,
and thank you for supporting better quality audio.

Support
For product support, please visit:
https://www.maat.digital/support
Would you like to help MAAT? Help support MAAT by sharing the
love…Like us on Facebook:
https://www.facebook.com/maatdigital/
and Twitter:
https://twitter.com/maat_digital

and tell your friends about us!

Credits
Product Concept, Design & Management:
OMas
Programming:
Adam, Agent of Shield
UI:
Friedemann “I love it!” Tischmeyer
Thanks Dr. Christoph Musialik for his contributions in shaping the
underlying engine on which LINpro is based. Thanks also to all the
scientists who, 30 years ago, performed the fundamental research work
that led to the development of all the sophisticated dither and noise
shaping theories on which later investigators built, including ours. In
turn, these pioneers paved a path to much better audio quality at shorter
word lengths: Gerzon, Craven, Lipshitz, Vanderkooy, Wannamaker, and
many others who have continued their investigations into this topic.

Page 23 of 37
Appendix One — The Math
For a deep dive into the basic theory of dithering, look no further than Robert A. Wannamaker’s 2003 Doctor-
al thesis entitled, “The Theory of Dithered Quantization”. A PDF version is available for your profound elucida-
tion here:
http://www.robertwannamaker.com/writings/rw_phd.pdf

Appendix Two — The Curves

Though our understanding of how dither is perceived by the human auditory system has evolved over time,
the fundamentals have not and were first described in the modern sense in 1961. I was first exposed to dither
as a high schooler, where the gas gauge in my car required mechanical dither, aggressive tapping on the gauge
face, to yield an accurate reading. Later, I dove in deep while working for Sonic Solutions, manufacturer of the
first professional DAW. Thank you, James A. Moorer for that!
For your deep dive, I suggest yuou start with, of all things, the Wikipedia page:
https://en.wikipedia.org/wiki/Dither
As a follow up, here are some loosely related additional writings from prominent audio practitioners you might
find of interest…
Eelco Grimm once wrote:
“What’s the difference between genuine dither and noisy lower bits?
The amplitude probability density [or power distribution versus frequency — Omas]. ‘Genuine’ dither has a
probability density function from a particular, rather restricted family. Sripad and Snyder proved that the
Fourier transform of the PDF (Probability Density Function) of the dither must go to zero at periodic distanc-
es from the origin. Practically, this means that proper dither is the sum of one or more random signals having
rectangular PDFs of width Q/2 (where) Q is the LSB size. The number of that are summed (which) determines
how many moments of the truncation error are de–correlated from the input signal. We use two in audio
because that de–correlates the power (as well as the mean) of the truncation error. Adding two independent
rectangular PDF dither components yields a triangular PDF (TPDF) dither signal.
Arbitrary noise doesn’t meet these requirements. It still helps, but if you listen to only the truncation error
signal, you’ll hear traces of the original program.”
Bob Katz once wrote:
“My usual solution to ‘cover up the distortion’ problem is to use one of the many different flavors of dither I
have available to me. One of them usually does better than others. One just has to be aware of the tradeoff in
apparent low level ambience and resolution. You never get something for nothing…the hotter the dither, sure,
you might hear less distortion and get a ‘smoother’ sound, but the low level ambience slowly leaches away.”
In response to Bob, Daniel Weiss wrote:
“There is only <em>one</em> right dither amplitude for any given dither flavour. Dither is not ‘covering’
distortion, it simply makes the quantization error to be uncorrelated from the audio signal and thus there is no
quantization distortion anymore - provided the dither level is high enough. Dither level may be higher than
necessary, but that does not ‘cover’ more distortion, it simply adds more noise.
I don’t see why manufacturers make the dither level variable by the user. [We at MAAT removed that option
in LINpro and LINone] Makes no sense - except maybe when the signal already has some noise floor which is

Page 24 of 37
 suitable for dithering, i.e. a relatively high noise floor.” [A noisy ADC would be a good example of that; con-
firmed by David Josephson & Jim Johnston. Johnson adds that, “Yes, delta-sigma convertors have relevant
dither issues.” — OMas]
Marc Lindahl once wrote:
“If you have ‘shaped’ dither, then you can run into problems with compromised headroom…Also, ‘perceptu-
al’ shapes can go from imperceptible to perceptible if ‘stacked’…other than that there’s probably more danger
from stacking quantization noise, since it’s kinda like harmonic distortion, and can quickly emerge from the
‘mud’…”

Appendix Three — The Curves

These are the same illustrations that appear within the plug–in. They are not MATLAB simulations, they are
actual plots generated by LINpro’s optimized dither engine. Notice that, in all but the first graph, the -100dBFS
input signal or excitation (the vertical spike at 997is preserved even though the word length has been reduced
to 16 bits.
A 16–bit word can only encode or store 96dB of dynamic range. As mentioned earlier, any information quiet-
er than -96dBFS should be lost when reducing the word length to 16. The first plot, an empty one, shows that
all information residing in the 17th through 24th bit in a 24–bit word is gone without proper dithering. For all
other illustrations in this appendix, the input signal is preserved.

Figure A1 — No dither, no noise shaping = no low level information preserved

Page 25 of 37
 Figure A2 — Rectangular dither, no noise shaping

Figure A3 — Rectangular dither, NS1

Page 26 of 37
Figure A4 — Rectangular dither, NS2

Figure A5 — Rectangular dither, NS3

Page 27 of 37
 Figure A6 — Triangular dither, no noise shaping

Figure A7 — Triangular dither, NS1

Page 28 of 37
Figure A8 — Triangular dither, NS2

Figure A9 — Triangular dither, NS3

Page 29 of 37
 Figure A10 — Triangular dither, PS1

Figure A11 — Triangular dither, PS2

Page 30 of 37
Figure A12 — Triangular dither, PS3

Figure A13 — Triangular dither, PS4

Page 31 of 37
 Figure A14 — High passed triangular dither, no noise shaping

Figure A15 — High passed triangular dither, NS1

Page 32 of 37
Figure A16 — High passed triangular dither, NS2

Figure A17 — High passed triangular dither, NS3

Page 33 of 37
 Figure A18 — High passed triangular dither, PS1

Figure A19 — High passed triangular dither, PS2

Page 34 of 37
Figure A20 — High passed triangular dither, PS3

Figure A21 — High passed triangular dither, PS4

Page 35 of 37
Page 36 of 37
About This Manual
This manual was written in Adobe InDesign 17.3, and is set in Robert Slimbach’s Minion Pro and Myriad Pro.
The cover page is set in Aldo Novarese’s modernist geometric Eurostyle. O.A. Masciarotte supplied the book
To Serve & Groove, from which some material was adapted for this LINpro user manual.
This documentation, as well as the software described in it, is furnished under license and may only be used or
copied in accordance with the terms of such license. The information in this manual is furnished for informa-
tional use only, is subject to change without notice, and should not be construed as a commitment by MAAT
Incorporated. MAAT Incorporated assumes no responsibility or liability for any errors or inaccuracies that
may appear in this document.
Except as permitted by such license, no part of this publication may be reproduced, stored in a retrieval sys-
tem, or transmitted, in any form or by any means, electronic, mechanical, recording, or otherwise, without the
prior written permission of MAAT Incorporated.

Notice of Liability
The author and publisher have made every effort to ensure the accuracy of the information herein. However,
the information contained in this document is provided without warranty, either express or implied. Nei-
ther the authors, MAAT Incorporated, nor its dealers or distributors will be held liable for any damages to be
caused either directly or indirectly by the instructions contained in this book, or by the software or hardware
products described herein.
MAAT INCORPORATED MAKES NO WARRANTIES, EXPRESS OR IMPLIED, INCLUDING WITHOUT
LIMITATION THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTIC-
ULAR PURPOSE, REGARDING THE APPLE SOFTWARE. MAAT INCORPORATED DOES NOT WAR-
RANT, GUARANTEE, OR MAKE ANY REPRESENTATIONS REGARDING THE USE OR THE RESULTS
OF THE USE OF THE MAAT INCORPORATED SOFTWARE IN TERMS OF ITS CORRECTNESS, AC-
CURACY, RELIABILITY, CURRENTNESS, OR OTHERWISE. THE ENTIRE RISK AS TO THE RESULTS
AND PERFORMANCE OF THE MAAT INCORPORATED SOFTWARE IS ASSUMED BY YOU. THE EX-
CLUSION OF IMPLIED WARRANTIES IS NOT PERMITTED BY SOME STATES. THE ABOVE EXCLU-
SION MAY NOT APPLY TO YOU.
IN NO EVENT WILL MAAT INCORPORATED, ITS DIRECTORS, OFFICERS, EMPLOYEES, OR AGENTS
BE LIABLE TO YOU FOR ANY CONSEQUENTIAL, INCIDENTAL, OR INDIRECT DAMAGES (INCLUD-
ING DAMAGES FOR LOSS OF BUSINESS PROFITS, BUSINESS INTERRUPTION, LOSS OF BUSINESS
INFORMATION, AND THE LIKE) ARISING OUT OF THE USE OR INABILITY TO USE THE MAAT
INCORPORATED SOFTWARE EVEN IF MAAT HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH
DAMAGES. BECAUSE SOME STATES DO NOT ALLOW THE EXCLUSION OR LIMITATION OF LIA-
BILITY FOR CONSEQUENTIAL OR INCIDENTAL DAMAGES, THE ABOVE LIMITATIONS MAY NOT
APPLY TO YOU.

©2019-2022 MAAT Incorporated — All rights reserved.

MAAT, LINpro, LINsurround and the MAAT logo are trademarks of MAAT Incorporated. All other company
or product names are either trademarks or registered trademarks of their respective owners.
220622-2.11
Page 37 of 37